Self-learning vehicle performance optimization

Patent metadata

Field	Value
Publication number	US-11535274-B2
Application number	US-202016831364-A
Country	US
Kind code	B2
Filing date	Mar 26, 2020
Priority date	Mar 26, 2020
Publication date	Dec 27, 2022
Grant date	Dec 27, 2022

Abstract

Provided herein is a system of a vehicle that comprises one or more sensors, one or more processors, and memory storing instructions that, when executed by the one or more processors, causes the system to perform: selecting a trajectory along a route of the vehicle; predicting a trajectory of another object along the route; adjusting the selected trajectory based on a predicted change, in response to adjusting the selected trajectory, to the predicted trajectory of the another object, the predicted change to the predicted trajectory of the another object being stored in a model; determining an actual change, in response to adjusting the selected trajectory, to a trajectory of the another object, in response to an interaction between the vehicle and the another object; updating the model based on the determined actual change to the trajectory of the another object; and selecting a future trajectory based on the updated model.

First claim

The invention claimed is:

1. A system of an autonomous or semi-autonomous vehicle comprising:
one or more sensors;
one or more processors; and
a memory storing instructions that, when executed by the one or more processors, causes the system to perform, without intervention from a human:
selecting a trajectory from potential trajectories along a route to be travelled by the autonomous or semi-autonomous vehicle;
predicting a trajectory of another object along the route;
adjusting the selected trajectory based on a current traffic density, a current lighting condition, and a predicted change, in response to adjusting the selected trajectory, to the predicted trajectory of the another object, the predicted change to the predicted trajectory of the another object being stored in a machine learning model and being based on instances of historical data of previous changes;
determining an actual change, in response to adjusting the selected trajectory, to the trajectory of the another object, based on a difference between the predicted trajectory of the another object and an actual trajectory of the another object in response to an interaction between the autonomous or semi-autonomous vehicle and the another object;
updating the machine learning model based on the determined actual change to the trajectory of the another object, the updating comprising:
determining weights, corresponding to the respective instances of the historical data, of the machine learning model, wherein the weights are based on degrees of similarity between:
previous lighting conditions corresponding to the historical data, and
the current lighting condition during which the actual change took place; and

adjusting the predicted change to the predicted trajectory of the another object according to a weighted average of the predicted change to the predicted trajectory of the another object and the actual change to the trajectory of the another object, the weighted average being based on adjusting the weights to reduce respective deviations between the predicted change to the predicted trajectory of the another object and the actual change to the predicted trajectory of the another object; and

selecting a future trajectory based on the updated machine learning model.

2. The system of claim 1 , wherein the adjusting the selected trajectory comprises:
adjusting the selected trajectory to an updated trajectory having a lowest predicted travel time and in which the predicted change to the predicted trajectory of the another object is within a predetermined range.

3. The system of claim 2 , wherein the updating the model comprises:
determining whether the actual change to the trajectory of the another object is within the predetermined range;
in response to determining that the actual change to the trajectory of the another object is within the predetermined range, determining a resulting change to the adjusted selected trajectory of the autonomous or semi-autonomous vehicle based on the actual change to the trajectory of the another object; and
in response to determining that the actual change to the trajectory of the another object is outside the predetermined range, adjusting the predicted change to the trajectory of the another object based on the actual change to the trajectory of the another object.

4. The system of claim 1 , wherein the predicted change to the trajectory of the another object comprises a predicted change in a velocity or an acceleration of the another object.

5. The system of claim 1 , wherein the adjusting the selected trajectory further comprises adjusting the selected trajectory based on a predicted change to a level of equilibrium of the autonomous or semi-autonomous vehicle and a trajectory of a second object predicted to contact the another object, in response to the predicted change to the trajectory of the another object.

6. The system of claim 1 , wherein the trajectory comprises a starting pose, a velocity at different points along the route, an acceleration at different points along the route, an orientation at different points along the route, and a final pose of the autonomous or semi-autonomous vehicle.

7. The system of claim 1 , wherein the adjusting the selected trajectory comprises adjusting the selected trajectory while maintaining a minimum distance, at a given time, between the adjusted selected trajectory and the trajectory of the another object.

8. The system of claim 1 , wherein the trajectory comprises four knot points that define a cubic spline that specifies a steering angle of the autonomous or semi-autonomous vehicle.

9. The system of claim 1 , wherein the selecting of the trajectory is based on a friction of a tire of the autonomous or semi-autonomous vehicle on a road surface.

10. The system of claim 1 , wherein the adjusting of the trajectory of the autonomous or semi-autonomous vehicle comprises straddling of the obstacle such that the autonomous or semi-autonomous vehicle passes over the obstacle and the obstacle is temporarily under an underbody of the autonomous or semi-autonomous vehicle.

11. The system of claim 1 , wherein the adjusting the selected trajectory comprises permitting a second vehicle to merge onto a same lane occupied by the autonomous or semi-autonomous vehicle.

12. The system of claim 1 , wherein the weights are based on respective degrees of similarity between a previous road type or previous road quality and a current road type or a current road quality.

13. A method implemented by a computing system including one or more processors and storage media storing machine-readable instructions, wherein the method is performed using the one or more processors of an autonomous or semi-autonomous vehicle, the method being performed without invention from a human and comprising:
selecting a trajectory from potential trajectories along a route to be travelled by the autonomous or semi-autonomous vehicle;
predicting a trajectory of another object along the route;
adjusting the selected trajectory based on a current traffic density, a current lighting condition, and a predicted change, in response to adjusting the selected trajectory, to the predicted trajectory of the another object, the predicted change to the predicted trajectory of the another object being stored in a machine learning model and being based on instances of historical data of previous changes;
determining an actual change, in response to adjusting the selected trajectory, to the trajectory of the another object;
updating the machine learning model based on the determined actual change to the trajectory of the another object, the updating comprising:
determining weights, corresponding to the respective instances of the historical data, of the machine learning model, wherein the weights are based on degrees of similarity between:
previous lighting conditions corresponding to the historical data, and
the current lighting condition during which the actual change took place; and

adjusting the predicted change to the predicted trajectory of the another object according to a weighted average of the predicted change to the predicted trajectory of the another object and the actual change to the trajectory of the another object, the weighted average being based on adjusting the weights to reduce respective deviations between the predicted change to the predicted trajectory of the another object and the actual change to the predicted trajectory of the another object; and

selecting a future trajectory based on the updated machine learning model.

14. The method of claim 13 , wherein the adjusting the selected trajectory compr

Assignees

• Pony Ai Inc

Inventors

• Dingli Robert
• Diehl Peter G
• Xiao Sinan

Classifications

• B60W50/0098

  Details of control systems ensuring comfort, safety or stability not otherwise provided for · CPC title

• B60W2556/50

  of positioning data, e.g. GPS [Global Positioning System] data · CPC title

• B60W2555/20

  Ambient conditions, e.g. wind or rain · CPC title

• G08G1/166

  for active traffic, e.g. moving vehicles, pedestrians, bikes · CPC title

• G08G1/0112

  from the vehicle, e.g. floating car data [FCD] · CPC title